Journal of Oral Rehabilitation 2004 31; 972978

Morphometric relationships between tooth and face shapes

H.-B. LINDEMANN*, C. KNAUER & P. PFEIFFER*

*Department of Prosthetic Dentistry, School of Oral

Medicine, University of Cologne, Cologne, Germany, Department of Computer Science, Institute of Computer Science, Free University of Berlin,
Berlin, Germany

SUMMARY The shape of a patients face is commonly

used as a reference to select the shape of the
maxillary central incisors in edentulous patients.
The validity of this relationship has not been
proved. The objective of this clinical study was to
determine whether a relationship exists between
maxillary central incisors and face shapes. Casts
were made of the maxillas of 50 men and 50 women.
A standardized digital photographic procedure was
used to record frontal views of each subjects face
and of the maxillary central incisors of the dental
casts. The shapes of the maxillary central incisors
were compared with the face forms. Shape matches
were evaluated according to their Hausdorff distance (HDD). The function h(A,B) is called the
directed HDD from shape A to shape B (this function is not a true distance). It reflects the distance of
the point of shape A that is farthest from any point
of shape B and vice versa. The similarity of both
shapes is given as a non-negative number. The value

Introduction
Reconstruction of anterior teeth can be a challenge in
dental prosthetics. By the end of the 19th century
attempts had been made to find individual features in
the patient that might provide clues to the shape of the
maxillary anterior teeth. In 1911, specific relationships
between face and tooth form were described and a
classification of maxillary central incisor shapes was
stated (1), followed by various studies, which tried to
find other correlations (25). Metric correlations
between the nose width and the width of the lateral
and central incisors were postulated (6), and it was
proved that the distance between the maxillary canines
(intercanine distance) was similar to the width of the
2004 Blackwell Publishing Ltd

00 indicates that the figures are identical (after

scaling and shifting). Higher values indicate that
shapes differ more substantially. Significant differences on the 5% level were calculated using the nonparametric MannWhitney U and KruskalWallis
tests. The face shape from the chin margin to the
eyebrow line (superior edges of the eyebrows)
produced a better match than the one from the
chin to the hairline (P < 00001). On average, the
maxillary central incisors displayed a variability
(0084
0028) that was higher by a factor of 19
than the face shapes (chin margin to the eyebrow
line, 0045
0015). In the interindividual comparison, the shapes of the maxillary central incisors of
women displayed a significantly smaller HDD than
the ones of the men (P < 00001).
KEYWORDS: tooth and face shapes, maxillary central
incisors, Hausdorff distance
Accepted for publication 8 December 2003

nose (7). The width of the maxillary central incisors was

found to be similar to the bizygomatical distance
divided by 16 (8) and it was stated that maxillary
incisors of women were smaller than those of men (9).
Several attempts have been made to describe a
relationship between the shape of maxillary incisors
and the face form. Only a limited number of studies on
this subject have been published over the past 20 years
(1015). Some studies postulated that the face form was
helpful to select the tooth form (16, 17), whereas others
did not find shape relationships between face and tooth
shape (5, 18). When the outline forms of both maxillary
central incisors were compared with face forms in
70 patients (18), it was found, that more than twothirds of the patients showed no similarity between face

972

TOOTH AND FACE SHAPES

form and incisor tooth form. To date, comparing face
shape to the inverted tooth form (square, tapered or
ovoid) based on Williamss classification is currently
advised (19, 20). In daily dental routine, shape cards are
normally used to select a specific set of teeth (17, 20).
The arrangement of shapes is frequently based on a
classification of three to four types (1, 2, 19).
The objective of this clinical study was to compare
tooth and face form based on the hypothesis that a
relationship exists between maxillary anterior teeth and
face shape.

Materials and methods

Fifty men (age range: 1874 years, mean: 458

177 years) and 50 women (1979 years, 432

168 years) from a private dental practice (Steinheim,
Germany) were selected for the study. Only patients
who still had a complete dentition in the maxilla and
mandible, without any enamel dysplasia and crown
reconstructions, veneers, etc., and whose tooth widths
had not been changed by stripping or similar methods
were selected. In addition, patients with little or no
incisal wear were included. Only patients with hairlines
that could be seen in the frontal projection were
selected.

Determination of facial shape

After palpation, the mental protuberances were marked
with a water-soluble pen on the skin of each patient
(Fig. 1a). The patients wore a hairband so that the
hairline and the forehead could be seen. The distance
between the markings was measured with a calliper
and a slide gauge.* A frontal photograph of each patient
was taken using a digital camera (DKC ID-1).

Determination of the tooth shape

The tooth shapes should be regarded as two-dimensional, orthogonal projections (Fig. 2a,b) (13). Irreversible hydrocolloid impressions of the maxillary anterior
teeth were made using partial impression trays
(RimLock
). These impressions were used to fabricate

*Dentaurum, Ispringen, Germany.

Sony, Tokyo, Japan.

Blend-a-med, Mainz, Germany.

Dentsply, York, PA, USA.

the corresponding plaster casts (Moldano
). For

enhanced contrast, a circular marking around each
anterior tooth was prepared with a lead pencil (Fig. 2a).
The casts were then measured and photographed. To
obtain an orthogonal projection type, each tooth was
photographed individually with the digital camera
(DKC ID-1).
For digital surveying of the face and tooth photographs, the images were transferred to a Macintosh
PowerBook** via a flash card or via the onboard
Small Computer Systems Interface (SCSI) controller to
a Power Macintosh 7600/120,** so that they could be
measured and edited (Figs 1c,d and 2b). For this
purpose, the freeware software program NIH Image
V.1.59 were used. The distance between the mental
protuberances was used to calibrate the digital frontal
picture (Fig. 1b). The data from the in vivo measurement were supplied for the distance between the
mental protuberances on the digital image.
Editing was carried out using the following steps.
First, the picture was converted into grey scale (Figs 1a
and 2a). Secondly, the projection type was selected
manually with use of the lasso function of a graphics
program (Figs 1a,b and 2b). Thirdly, all non-relevant
data were deleted and the contour shape was generated
as a black and white bitmap file (Figs 1c,d and 2c) and
then converted to a vector graphic.
The face forms were determined from the hairline to
the chin margin (group hairline/chin: HC; Fig. 1a) as
well as from the eyebrow line (superior edges of the
eyebrows) to the chin margin (group eyebrow line/
chin: EC; Fig. 1b).

Mathematical basis of the comparison

algorithm
The similarity of contour shapes was determined using
the Hausdorff distance (HDD) method (2123). The
HDD measures the degree of mismatch between two
shapes A and B. Before determining HDD, scaling and
shifting of shape B are applied to adapt shape B to
shape A for maximum matching. The function h(A,B) is
called the directed HDD from shape A to shape B (this
function is not a true distance and has no SI unit). It

Heraeus Kulzer, Dormagen, Germany.

**Apple, Cupertino, CA, USA.

SanDisk, Sunnyvale, CA, USA.

National Institutes of Health, Bethesda, MD, USA.

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978

973

974

H . - B . L I N D E M A N N et al.

Fig. 1. The orthogonal projection of

the patients face was selected
manually with use of the lasso
function of a graphics programme.
The face forms were determined
from the hairline to the chin margin
(a: group hairline/chin: HC) as well
as from the eyebrow line (superior
edges of the eyebrows) to the chin
margin (b: group eyebrow line/chin:
EC). The distance between the
mental protuberances (MP) was used
to calibrate the digital frontal picture
(a, b). The contour shapes of the
patients face were generated as a
black and white bitmap file (c: group
HC; d: group EC).

reflects the distance of the point of shape A that is

farthest from any point of shape B and vice versa
(Fig. 3). Details of the mathematical basis of the HDD
can be reviewed in several studies (2123). As there are
no efficient algorithms known to compute this measure
exactly, an approximation algorithm is used (22). The
similarity (HDD) of both shapes is given as a nonnegative number (no SI unit). The value 00 indicates
that the figures are identical (after scaling and shifting),
whereas higher values indicate that the shapes differ
more substantially.
The comparisons of the shapes (Computational
Geometry Algorithms Library) (24) were carried out
using the operating system Linux.

To determine the best facial shape fit to the silhouette

of the maxillary central incisors (Fig. 3), the HDDs of
each maxillary central incisor in all 100 patients were
measured using the different facial shapes (HC, EC) and
the average HDD of the maxillary central incisors of
each patient was calculated.
In order to examine the hypothesis of a three- or
four-class pattern of the tooth and facial forms, the
distribution of the HDD values in the interindividual
comparisons of tooth and face shapes were analysed.
For verification of the statement that men and
women tooth shapes are different, tooth shapes of
maxillary left and right central incisors were compared
within two groups of 50 men and women each.

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978

TOOTH AND FACE SHAPES

Fig. 2. For enhanced contrast, a circular marking around the maxillary

central incisor was prepared with a
lead pencil (a). The orthogonal projection of the tooth was selected
manually with use of the lasso function of a graphics programme (b).
The contour shape of each anterior
tooth was generated as a black and
white bitmap file (c).

Face outline
Tooth outline

Hausdorff distance

Fig. 3. The similarity of contour shapes was determined using the

Hausdorff distance (HDD). It identifies the point that is farthest
from any point of B, and measures the distance from B to its
nearest neighbour in B. Illustration of the HDD between the face
and the tooth outlines.

Another point of interest was to investigate the theory

that mens teeth have a larger volume than those of
women (9). As the tooth shapes were analysed using
two-dimensional projections, no direct information on
the volume of the teeth could be drawn from the images,
so the area of the tooth silhouette was used as a criterion
of evidence. The area of each maxillary central incisor
was determined by calculating the percentage of the
black content of the tooth shape projections.
Significant differences on the 5% level were calculated using the non parametric MannWhitney U and
KruskalWallis (maxillary central incisor versus face
shape) tests.

Results
Table 1 shows the results of the HDDs of intra- and
interindividual comparisons. The study revealed that

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978

975

H . - B . L I N D E M A N N et al.

Mean

Standard deviation

Comparisons of the maxillary central incisors

Left/right
Interindividual
Men
Women

0065
0084
0085
0081

0023
0028
0025
0024

Comparisons of maxillary central incisor

and face shapes
Eyebrow line/chin margin*
Eyebrow line/chin margin
Hairline/chin margin
Men (eyebrow line/chin margin)*
Women (eyebrow line/chin margin)*

0075
0084
0108
0080
0071

0030
0022
0027
0033
0028

Table 1. Hausdorff distances of

intra- and interindividual comparisons

*Without consideration of the incisal edge.

With consideration of the incisal edge.

180
160
140
120
100

100
90
80
70
60

the tooth shapes of maxillary left and right central

incisors of each patient are more similar than the tooth
shapes of the respective maxillary incisors of different
patients, i.e. the HDD is smaller (P < 00001). However,
maxillary left and right central incisors are not
completely identical.
The comparisons of the tooth shapes with different
face shapes are displayed in Table 1. The face shapes of
the EC group produced a better match than the one of
the HC group (P < 00001).
On average, the maxillary central incisors displayed a
variability (0084
0028) that was higher by a factor
of 19 than the face shapes (group EC, 0045
0015).
From the values calculated for each tooth pair no
relationship can be deduced.
The HDD distributions of the maxillary central
incisors and the face shapes (EC) are shown in
histograms (Figs 4 and 5). The curves revealed a slightly

50
40
30
20
10
0
0.00

0.05

0.10

0.15

0.20

Hausdorff distance

Fig. 5. Distribution of the Hausdorff distances of the maxillary

central incisors in interindividual comparisons.

shifted Gaussian distribution in relation to the values of

their arithmetical mean and standard deviations.
The values for the average HDD and the standard
deviation are included in Table 1. In the interindividual
comparison, the shapes of the maxillary central incisors
of women displayed a significantly smaller HDD than
the mens (P < 00001). Womens tooth shapes of
maxillary central incisors exhibited a smaller variation
range than those of men. HDDs of the maxillary central
incisors without consideration of the incisal edge

976

80

Table 2. Height and width of maxillary central incisors of men

and women

60
40
20
0
0.00

Height (cm)
0.05

0.10

0.15

0.20

Width (cm)

Mean Standard deviation Mean Standard deviation

Hausdorff distance

Fig. 4. Distribution of the Hausdorff distances of the face shapes

(eyebrow line/chin margin) in interindividual comparisons.

Men
Women

109
104

014
011

091
089

006
008

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978

TOOTH AND FACE SHAPES

compared with face shapes (EC) revealed significant,
sex-specific differences (Table 1, P 0038).
The height of maxillary central incisors displayed
significant sex-specific differences (P 0023, Table 2).
The maxillary central incisors of men are longer. The
average deviation in height is c. 5%. There were no
significant differences between the width of maxillary
central incisors of men or women (P 0088, Table 2).

Discussion
Fifty men and 50 women older than 18 years were
examined. The two gender-based groups, however, did
not include subjects with extremely worn maxillary
anterior teeth, as they occur as a result of parafunction.
The methods used for calibration and acquisition of
specific data were examined in various pre-studies and
revealed only minor deviations of 1%.
Precise areas were obtained by the bitmap graphics of
the orthogonal projections of the maxillary anterior teeth
and the face shapes. Furthermore, using the method of
determination of the smallest-directed HDD (22), a nonnegative number was obtained to describe the relationship between all shapes. The reliability of the method was
confirmed for the shape matching of binary images (21)
as well as for the automatic face recognition of an
HDD-based measure (23).
This study revealed, that a classification into three or
four types could not be made in relation to the orthoradial contour shapes of maxillary left and right central
incisors and of the faces. A classification into three types
was first postulated by Williams (1, 2). Other authors
(3, 4) confirmed Williamss classification; the theory of
shapes was even extended to four types of shape (19). In
the present study, a slightly shifted Gaussian distribution
of the HDD values of the maxillary central incisors and
the face shapes was obtained (Figs 4 and 5). The type
theory (1, 2, 16) can be refused, because shape comparisons of the same type should provide better matches and
thus smaller HDDs. This should become obvious
throughout the accumulation of values in various peaks
not in a Gaussian distribution. Thus, further adherence to
this theory would not be useful.
The silhouettes of the faces below the eyebrow level
exhibited a correlation with those of the maxillary
central incisors. In this respect, the results that were
found correspond with those of Williams (1, 2) and
other authors (3, 16, 19). In the orthogonal projection,
the shape of the maxillary central incisors revealed a

variation that is twice as high as the one of the lower

half of the faces, but the face contour below the
eyebrows can limit the variation of shapes for the
cervical contour of the maxillary central incisors and
thus provide information on the tooth shape.
It is conceivable that manufacturers could offer shape
cards in the form of software. The shape of the artificial
teeth offered would be stored in the program.
A systematic comparison of the lower face contour
with artificial tooth shapes would provide corresponding HDD numbers. The tooth shape with the lowest
HDD would be selected.
For long-term patient treatment, it would be useful to
save digital photographs of casts of the maxillary
anterior teeth in the dental practice software. In case
of loss of the anteriors, the graphics could be evaluated
and using software shape cards the shapes that
offered the best match could be selected. This procedure
would offer the highest precision of reproduction and
could be helpful in particular for reconstruction work in
patients with trauma in the anterior maxillary region
or rapidly progressing juvenile periodontitis.
Based on a mathematical approach on shape matching, the results of the present study can be helpful for
selecting the tooth form for edentulous patients. This
study refuted the three to four type theory, revealed
that the face contour below the eyebrows can provide
information on the tooth shape and showed sex-specific
differences in the height of maxillary central incisors.

Conclusions
Within the limitations of this study, the following
conclusions were drawn:
1 Neither maxillary central incisor nor face shapes
could be classified in a three- or four-class pattern.
2 The height of maxillary central incisors displayed
significant sex-specific differences. The maxillary central incisors of men were longer. There were no
significant differences between the width of maxillary
central incisors of men or women. Womens tooth
shapes of maxillary central incisors exhibited a smaller
variation range than those of men.
3 Maxillary central incisors displayed a variability that
was higher by a factor of 19 than the face shapes (chin
margin to the eyebrow line). Maxillary central incisors
produced a better match with the face shape from the
chin margin to the eyebrow line than from the chin to
the hairline.

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978

977

978

H . - B . L I N D E M A N N et al.

References
1. Williams JL. Esthetics and anatomical basis of dental prothesis. Dent Cosmos. 1911;53:11.
2. Williams JL. New evidence of mans relationship to the
anthropoid apes. J Dent Res. 1928;8:805.
3. Stein R. Williams classification of anterior tooth forms. J Am
Dent Assoc. 1936;23:1512.
4. Kern BE. Anthropometric parameters of tooth selection.
J Prosthet Dent. 1967;17:431.
5. Bell RA. The geometric theory of selection of artificial teeth: is
it valid?. J Am Dent Assoc. 1978;97:637.
6. Gerber A. Complete dentures (VIII) creative and artistic tasks
in complete prosthodontics. Quintessence Int. 1975;6:45.
7. Lee J. Dental Aesthetics. Bristol: Wright; 1962.
8. Landa LS. Practical guidelines for complete denture esthetics.
Dent Clin North Am. 1977;21:512.
9. Frush JP, Fisher RD. How dentogenic restorations interpret
the sex factor. J Prosthet Dent. 1956;6:441.
10. Brodbelt RH, Walker GF, Nelson D, Seluk LW. Comparison of
face shape with tooth form. J Prosthet Dent. 1984;52:588.
11. LaVere AM, Marcroft KR, Smith RC, Sarka RJ. Denture tooth
selection: an analysis of the natural maxillary central incisor
compared to the length and width of the face. Part I. J Prosthet
Dent. 1992;67:661.
12. LaVere AM, Marcroft KR, Smith RC, Sarka RJ. Denture tooth
selection: an analysis of the natural maxillary central incisor
compared to the length and width of the face: part II.
J Prosthet Dent. 1992;67:810.
13. Sellen PN, Jagger DC, Harrison A. Computer-generated study
of the correlation between tooth, face, arch forms, and palatal
contour. J Prosthet Dent. 1998;80:163.
14. Sellen PN, Jagger DC, Harrison A. The correlation between
selected factors which influence dental aesthetics. Prim Dent
Care. 1998;5:55.

15. Robinson DL, Blackwell PG, Stillman EC, Brook AH. Planar
procrustes analysis of tooth shape. Arch Oral Biol.
2001;46:191.
16. Selg TE. Selecting and arranging teeth for full dentures. Dent
Lab Rev. 1982;57:17.
17. Sellen PN, Jagger DC, Harrison A. Methods used to select
artificial anterior teeth for the edentulous patient: a historical
overview. Int J Prosthodont. 1999;12:51.
18. Mavroskoufis F, Ritchi GM. The face-form as a guide for the
selection of maxillary central incisiors. J Prosthet Dent.
1980;43:501.
19. Seluk LW, Brodbelt RHW, Walker GF. A biometric comparison of face shape with denture tooth form. J Oral Rehabil.
1987;14:139.
20. Ibrahimagic L, Jerolimov V, Celebic A, Carek V, Baucic I,
Zlataric DK. Relationship between the face and the tooth
form. Coll Antropol. 2001;25:619.
21. Huttenlocher DP, Klanderman GA, Rucklidge WJ. Comparing
images using the Hausdorff distance. IEEE Trans Pattern Anal
Machine Intell. 1993;15:850.
22. Aichholzer O, Alt H, Rote G. Matching shapes with a reference
point. Int J Comput Geometry Appl. 1997;7:349.
23. Pujol A, Villanueva J, Alba JL. A supervised modification of
the Hausdorff distance for visual shape classification. Int J
Pattern Recognit Artif Intell. 2002;16:349.
24. Fabri A, Giezeman GJ, Kettner L, Schirra S, Schonherr S. On
the design of CGAL, the computational geometry algorithms
library. Software Pract Exper. 2000;30:1167.

Correspondence: Peter Pfeiffer, Kerpener St. 32, D-50931 Koln,

Germany.
E-mail: peter.pfeiffer@uni-koeln.de

2004 Blackwell Publishing Ltd, Journal of Oral Rehabilitation 31; 972978